1 For further information on the structure and availability of the Brown family of corpora see the homepage of ICAME, the International Computer Archive of Modern/ Medieval English, at http: / / helmer.aksis.uib.no/ icame.html. For information on additional corpus sources mentioned see David Lee’s “Bookmarks for Corpus-Based Linguists” at http: / / devoted.to/ corpora. The corresponding information on the DCPSE can be obtained from http: / / www.ucl. ac.uk/ english-usage/ projects/ dcpse/ research.htm. AAA - Arbeiten aus Anglistik und Amerikanistik Band 34 (2009) Heft 1 Gunter Narr Verlag Tübingen Ongoing grammatical change in spoken British English: real-time studies based on the DCPSE Barbara Klein 1. Introduction In the early 1990s, the completion of the Freiburg updates of the LOB and Brown corpora inaugurated a small boom in the corpus-based study of ongoing grammatical change in English in real time (see Mair 2006b: 12-35 for a discussion of methodological issues and Mair and Leech 2006 for a survey of important results). Given the composition of the “Brown family” of corpora, this research was necessarily restricted to the study of written English, and the extent to which diachronic developments observed in the written data reflected parallel trends in the spoken language remained controversial. The year 2006 saw the release of the Diachronic Corpus of Present-Day Spoken English (DCPSE), compiled by the Survey of English Usage at the University College London. The DCPSE is an 800,000 word corpus containing matching samples of spontaneous spoken British English from the London Lund Corpus (1958-77) and from the British component of the International Corpus of English (1990-92). For the first time, it is now possible to carry out “real time” studies on spoken data in roughly the same time span which is covered by the written Brown, LOB, Frown and F-LOB corpora. 1 The present paper uses the DCPSE to investigate three grammatical variables which have already been shown to be undergoing change in the written language: the s-genitive, the subjunctive, and a selection of modal Barbara Klein 30 2 Due to the F-LOB corpus still being under construction at that time, only F-LOBpress was available for her research. Therefore, the analyses in the other corpora were restricted to the press section to allow full comparability of data. 3 The relative frequency of nouns from LOB to F-LOB remained relatively stable; Mair et al. (2002) documented only a slight increase by 0.8% in the relevant text types. verbs. In order to facilitate comparability to previous work based on the Brown family of corpora, I selected the late 1950s (1958-1960) and the early 1990s (1990-1992). Since thirty years are too short to track a grammatical change running its course from its primary emergence to its completion, this paper should be considered as the study of recent episodes in ongoing processes of change. The overall aim is to shed light on the extent to which changes proceed along parallel lines in speech and writing. 2. The s-genitive Popular surveys on language change such as Barber (1964) or Potter (1969/ 75) have long argued that the frequency of the s-genitive is increasing in present-day English. Barber even claims a spread of the s-genitive to inanimate nouns, at the expense of semantically equivalent of-constructions (1964: 132). Even though these early studies made people aware of the issue, they are based on impressionistic observation only. Subsequent corpus-based research, however, has made clear that these impressions are not totally without foundation. Raab-Fischer (1995) investigates the press section of the LOB/ F-LOB and the Brown/ Frown corpora and shows that in newspaper language, 2 the s-genitive is increasing with all kinds of possessor nouns, however not necessarily with those nouns ranking highest on the gender scale. The of-construction is shown to be losing ground with all types of possessor nouns. On the basis of an extensive elicitation study, Rosenbach claims that the s-genitive is increasing in present-day English, and that this extension is more productive than has been assumed so far (2002: 161). Mair (2006a: 245-246), too, confirms the general rise in the frequency of s-genitives in press and academic writing. 3 However, he emphasizes that in his data the observed increase is not due to an increasing use of the s-genitive with inanimate nouns (which had been suggested as the responsible factor in other studies), but rather to a text-type specific desire in written genres to compress information. Given the relatively short period of investigation (c. 30 years), the statistical shifts are impressive. Mair and Leech (2006) as well as Leech and Smith (2006) report an increase in the frequency of s-genitives of 24.1%. Genitive-equivalent of-phrases, by contrast, decline by 23.4% (Mair and Leech 2006: 334). According to Leech and Smith, in present-day written British English, “the loss of of-genitives is very roughly commensurate with the gain of s-genitives” (2006: 197). Like Ongoing grammatical change in spoken British English 31 4 FRED (The Freiburg Corpus of English Dialects) contains dialectal speech by older speakers of British English, mainly recorded between 1970 and 1990. The CSAE (The Corpus of Spoken American English) consists of about 166,000 words of informal American English (Hinrichs and Szmrecsanyi, forthcoming). 5 S-genitives were retrieved through the grammatical tagging of the corpus. For the ofconstructions a Fuzzy Tree Fragment had to be created manually. 6 I limited the comparison to the data of 1958 to 1960 and the year 1990 for two reasons: first, it would have been far too laborious to extract the interchangeable of-constructions from the whole of the ICE-corpus (1990-1992), since the Fuzzy Tree Fragment identified 4,510 potential s-genitive equivalents. Second, the number of words in LUND 1958-1960 (37,225) almost exactly matches the number of words in ICE 1990 (40,819), which makes it an ideal basis for comparisons. 7 Log likelihood is a measure of statistical significance: a value of 3.84 or more equates with chi-square values >0.05; a value of 6.63 or more equates with chi-square values >0.01. 8 The column headed “Diff (%)” gives the increase (+) or decrease (-) in occurrences as a percentage of the frequency in the 1958-60 data. Raab-Fischer, Hinrichs and Szmrecsanyi (2007) base their work on the press section of the LOB/ F-LOB and Brown/ Frown corpora. Another paper (2008) includes two non-matching spoken corpora (FRED and the CSAE 4 ). The authors argue that the increase in the s-genitive in press texts “may well reduce, at least partly, to an increasingly powerful tendency to code thematic NPs with the s-genitive” (2007: 468). They explain the increase in the s-genitive in terms of “economization”, the use of a more compact coding option in writing (2007: 469), as was also suspected by Mair (2006a: 245). The question is whether grammatical change is documented at such massive rates in writing, are these changes also happening in speech? On the basis of the DCPSE, as I will show below, this question has to be answered in the negative. For the identification of semantically equivalent s-genitives and of-constructions 5 , I used the criteria listed in Hinrichs and Szmreczanyi (2007: 446f.), Kreyer (2003: 170) and Raab-Fischer (1995: 127). Table 1 presents the results and shows no discernible diachronic trend: Table 1: Genitives interchangeable with of-phrases in two subsamples of DCPSE DCPSE 1958-60 1990 6 Log likhd 7 Diff (%) 8 total n/ 10,000 total n/ 10,000 s-genitive 67 18.00 74 18.13 0 + 0.72 of-construction 88 23.64 78 19.11 1.88 - 19.17 s-gen 1958-60 vs. 1990: not significant; of-construction 1958-60 vs. 1990: not significant Barbara Klein 32 The low log likelihood values demonstrate that, contrary to expectations, there is no significant change in the use of the forms in spontaneous spoken British English within the thirty years considered. The frequency of s-genitives remains stable, showing none of the drastic increase observed in writing. The use of the of-construction is decreasing slightly, yet below the degree of statistical significance. It has been argued that the s-genitive is spreading to inanimate possessor nouns in the written register (cf. Barber 1964: 132f.). Whatever the merit of this explanation may be for written data, it is irrelevant for the spoken language. A look at the types of s-genitive possessors in the DCPSE reveals that there are no instances of inanimate possessors in the years considered for this study. In sum, my analysis has illustrated that there is no diachronic change in genitive usage in spoken English. On the other hand, the increase of s-genitives in written English is so well documented that it cannot be doubted. The explanation for this discrepancy is that we are most likely not dealing with a genuine grammatical change here (in the sense of a change in the underlying system of choices and options), but with a stylistic development or an evolution of text-type specific norms in some written genres. Colloquialization, the drift toward more oral modes of expression in writing, has been named as an explanation for the spread of s-genitives in writing (cf. Leech and Smith 2006: 197). However, the increase in locative, temporal (Raab-Fischer 1995: 126) or abstract inanimate possessors (Dahl 1971: 170) can hardly be explained by colloquialization, since the DCPSE clearly indicates that these forms are not very frequent in speech. Rather, the increasing frequency of the s-genitive reflects a growing demand for economy in language, and an increase in the information-density of some written genres. 2. The subjunctive Toward the beginning of the 20 th century, most grammarians considered the English subjunctive a moribund form on the verge of disappearing from the language completely (cf. Jespersen 1905: 205, or Fowler 1926: 595). In the second half of the century, a comeback was noticed, especially in American English (cf. Hirtle 1964, Jacobsson 1975). Övergaard (1995) is the first systematic diachronic study to look at the developments of the subjunctive in the 20 th century, using different text corpora. She discovers a significant increase in the mandative subjunctive from 1960 to 1990 in British English, which she believes is due to American influence. In contrast to Övergaard (1995), Hundt (1998) is not able to confirm such a drastic increase in sub- Ongoing grammatical change in spoken British English 33 9 LOB, Brown, F-LOB, Frown, WCNZE = Wellington Corpus of Written New Zealand English (1980s), ACE = Australian Corpus of English (late 1980s), BNC = British National Corpus (spoken part) 10 Since her results are not statistically significant due to low numbers, the proclaimed trends still need to be checked on the basis of larger corpora. 11 The verb be is an exception, because its base form be is distinct to all singular and plural present indicative forms. Past subjunctive were is only non-ambiguous in the first and third person singular. junctives comparing LOB to F-LOB. Her overall results for corpora representing different varieties, 9 however, suggest that the subjunctive is becoming less formal (1998: 167). More recent studies by Serpollet (2001), Mair and Leech (2006), and Leech and Smith (2006) are all exclusively based on the Brown family of corpora. Serpollet (2001) reports a definite increase in the mandative subjunctive in British English, accompanied by a decline in periphrastic should. 10 Mair and Leech (2006: 328) speak more cautiously of a “modest revival” of the present subjunctive in British English alongside a continuing decline of the were-subjunctive. Apart from Hundt’s (1998) analysis of a sample of the spoken part of the BNC, there are no corpus-based studies on the use of the subjunctive in spoken English, let alone diachronic accounts of recent developments. In how far the spoken language has been affected by the revival of the mandative subjunctive is thus an issue that still needs to be looked into. The following analysis of the DCPSE covers both the mandative and the past subjunctive and will also examine the formulaic subjunctive and the subjunctive in clauses of condition and concession. In the analysis of the mandative subjunctive, modal-verb periphrases and indicative forms will also be considered. Since the present subjunctive is morphologically distinct from the indicative only in the third person singular, many ambiguous forms are expected. 11 Table 2 shows the total number of subjunctives and their periphrastic and indicative alternants in subordinate clauses governed by mandative expressions: Barbara Klein 34 12 Since the periods considered do not represent exactly the same number of words, the frequency of hits was normalized per 10,000 word tokens. 13 Hundt (1998) reports a decline of the mandative use of should by 17.28%, whereas Serpollet (2001) arrives at 46.45%, both relying on LOB and F-LOB. Table 2: Distribution of the mandative subjunctive and its alternants in two sub-samples of the DCSPE DCPSE 1958-1960 1990-1992 Log likhd Diff (%) total n/ 10,000 12 total n/ 10,000 Subjunctive 0 0.00 2 0.05 0.34 (+ 100.00) should 5 1.34 11 0.26 7.10* - 80.56 Other modals 2 0.54 9 0.21 1.14 - 60.24 Indicative 1 0.27 45 1.07 3.01 + 297.55 Non-distinct forms 0 0.00 10 0.24 1.69 (+ 100.00) Total mandative 8 2.15 77 1.83 0.18 - 14.88 * should 1958-60 vs. 1990-92: significant at p>0.01 The fact that there are effectively only two genuine subjunctive forms in the 458,587 words analysed is a clear indication of its insignificance in speech: (1) You remember she had insisted that they only do a small incision under local anaesthetic <DI-B80_12> (2) When we got the request for help we suggested that we look not at general loans <DI-G02_102>. Since the verb in (1) is not subject to the normal sequence of tenses - which would require backshifting the verb to did - and since phrase (2) forms its negation without do-support, they cannot be indicative. However, the subjunctive in (2) is somehow debatable, because, according to the rule of grammar books, not has to precede the verb in genuine subjunctive constructions. The subjunctive may be “alive and kicking” (Serpollet 2001: 531) in written English; it is still moribund, though, in contemporary spoken British English. The DCPSE data furthermore reveals that the frequency of the periphrastic construction with the modal should is on the decrease by 80.56% (significant at p > 0.01) in spoken British English, too. As the literature does not agree on the scope of the decline between the written corpora LOB and F-LOB due to different counting procedures, 13 it can only be said that the decline of periphrastic should seems to be more pronounced in Ongoing grammatical change in spoken British English 35 14 Log likelihood of 4.30 15 normalized per 10,000 word tokens: 1958-60: 1.61; 1990-92: 0.09. spoken English than in writing. In the DCPSE, modals other than should are also on the decline; one can however not generalize from these data, since the results are not statistically significant. The results for the indicative and non-distinct verb forms are very interesting within the scope of the DCPSE, though not significant enough statistically to be used as a basis from which any broad trends can be drawn. The frequency of indicatives replacing subjunctives in mandative construction indeed increases by 297.55% from 0.27 in the late 1950s to 1.07 hits per 10,000 word tokens in the early 1990s. Statements of the type (3) are becoming increasingly common. The question is, however, whether the nondistinct forms in the DCPSE exemplified in (4) actually represent genuine subjunctives or indicatives. (3) The key issue is how do we insure that the local government survives the poll tax <DI-D14_119> (4) I suggest that you sort of write or speak with Grandma <DI-C05_299> It is generally agreed that ambiguous forms should be added to the numerically more important group, which in this case obviously are the indicatives. Consequently, the inclusion of the non-distinct forms yields an increase in the frequency of indicatives by 385.89% 14 (significant at p > 0.05). This result is in line with the study on written British English by Hundt (1998), who also claims that the indicative is gaining ground. The synchronically most striking finding is that the indicative is the most frequent substitute for the genuine subjunctive in spoken British English of the early 1990s. Periphrasis by modals is used in about one quarter of all cases, while the genuine subjunctive is distinctly the least frequent option in mandative clauses. The formulaic subjunctive turned out to be, as expected, a relatively rare form. Nevertheless, it occurred more often than the genuine subjunctive. Even though one cannot speak of a drastic decline due to low numbers, there is still a slight decrease in its usage: down from 6 instances in the late 1950s to 4 examples in the early 1990s. 15 The formulaic expressions used remain the same, ranging from simple be it to far be it, so be it and if need be. It looks as if these kinds of expressions are generally quite rare in spontaneous speech - probably due to their archaic nature. Barbara Klein 36 The DCPSE offers very few examples of the use of the subjunctive in clauses of condition and concession. Although there is not a single form in the late 1950s, five attestations can be found in the years 1990 to 1992. Three are triggered by the expressions if, whether and even though, and involve the verb be as in (5) If we‘re advancing, even though that be quite slow, quite different attitudes prevail <DI-I01_55>. The other two examples contain the verb come used in a conditional sense. The frequency of were-subjunctives has not changed dramatically in spoken English within the thirty years considered either. Against the background of a continuing decline in writing (cf. Mair and Leech 2006: 329), it could be expected that were would be losing ground in speech as well. In fact, there is a modest decline (by 16.07%), which, however, involves very low overall numbers and is thus not statistically significant. Interestingly, was is on the decline in hypothetical clauses as well, dropping by 53.3% (significant at p > 0.05). Therefore, the assumption that the decline in past subjunctives is counterbalanced by an increase in indicative was turned out to be not true for spontaneous spoken language. From a synchronic point of view, it is worth mentioning that indicative was is still the most frequent option in hypothetical contexts. In the years 1990 to 1992, 79.57% of all conditional and concessive statements in the first and third person singular preferred indicative was to the formal were-subjunctive. Consequently, Johansson and Norheim’s argument that the “were-subjunctive is clearly the dominant choice in hypothetical-conditional clauses” (1988: 34) did not turn out to be true for spontaneous speech of the early 1990s. It should be noted, moreover, that the main use of the were-subjunctive is in the formula as it were, which is used almost twice as much as the genuine past subjunctive in the years 1990 to 1992. Most of the 35 uses of as it were concentrate on the two subcorpora B (informal face-to-face conversation) and I (assorted spontaneous), where the expression serves as a type of discourse marker. The use of this form in spoken American English has not been investigated so far. Johansson and Norheim (1988: 34) were, however, able to demonstrate that as it were is much more frequent in LOB than in the Brown corpus. Whether as it were is a Briticism in speech as well is a question which can only be answered by a synchronic comparison of British and American spoken corpora. Summing up, then, and looking at the results from a slightly different angle by combining the numbers of should and all other modals as well as the indicatives and the non-distinct forms, these are the salient developments in the thirty-year period under review: Ongoing grammatical change in spoken British English 37 Table 3: Distribution of should + other modals and indicative + non-distinct forms in mandative clauses DCPSE 1958-1960 1990-1992 Log likhd Diff (%) total n/ 10,000 total n/ 10,000 Subjunctive 0 0.00 2 0.05 0.34 (+ 100.00) Should + other modals 7 1.88 20 0.47 7.64 - 74.76 Indicative + non-distinct forms 1 0.27 55 1.31 4.30 + 385.89 should + other modals 1958-60 vs. 1990-92: significant at p>0.01; indicative + nondistinct forms 1958-60 vs. 1990-92: significant at p>0.05 While, in terms of the subjunctive, one cannot speak of any developments, since it is virtually not used, the indicative is gaining ground extensively in spoken mandative clauses. One could even assume that the moderate loss in should and other modals is in parts compensated by the increase in indicatives. These results raise the question why the subjunctive is “alive and kicking” in written British English, but hardly used in spontaneous speech. First of all, as in the case of the genitive, style and formality seem to play a pivotal role. Previous studies have shown that the development in written British English is remarkable, because the revival of the formal subjunctive cannot be explained by a colloquialization of the written norm. If it is thus due to an Americanization of British writing, one should be aware of the fact that such a process is apparently not at work in spoken British English. Why is the subjunctive in speech not subject to Americanization? Probably, as it is still, even in American writing, a formal way of expressing commands or propositions, which is not typical of spontaneous speech. Övergaard (1995: 89) attributes the increase in mandative subjunctives to the impact of American texts after World War II. The present DCPSE analysis shows that this influence has not affected spoken language. It looks as though British writers are embracing the subjunctive, because it is an optional stylistic device, which offers variation in writing whenever a more formal option is more appropriate. Like spoken British English, spoken American English has not been subject to real-time studies due to the lack of comparable corpora. Even though the role of the subjunctive in written American English has been thoroughly investigated, the only available source for speculations on the use of the subjunctive in speech are elicitation studies such as Nichols (1987) or Algeo (1988). These studies suggest that the subjunctive Barbara Klein 38 16 SEU: Survey of English Usage spoken “Mini-Corpus”. 80,000 words from selected categories of the years 1959-65. ICE-GB: International Corpus of English spoken “Mini-Corpus”. 80,000 words from selected categories of the years 1990-92. 17 Including needn’t, the form is decreasing by 40.2% (Leech 2003: 228). in American speech is not as rare as in British English, however, since these studies are not empirical and merely based on questionnaires, a diachronic set of comparable spoken corpora of American English is definitely required to find out whether the two varieties are diverging or converging. Once again, we have seen that a diachronic development in writing does not necessarily entail a corresponding trend in speech. 4. Modal auxiliaries must, have to, (have) got to, need (to), shall As several corpus-based studies have shown, modality and the modal verbs have been “on the move” (Leech 2003) in the recent past, at least in written English (Leech 2003, 2004b; Mair and Leech 2006; Leech and Smith 2006). Unlike the s-genitive with inanimate nouns or the subjunctive, modals are extensively used in spoken English, which is why Leech (2003, 2004b) carried out a pilot study on a small subsection of what was to become the DCPSE. Even though these “mini-corpora” were too small to retrieve any statistically significant trends, it still became apparent that changes in the use of the modal auxiliaries should be much more pronounced in spoken English than in writing. Table 4 summarizes Leech’s results for the verbs studied also in the present paper: Table 4: Frequencies of modals and semi-modals in written and spoken BE (Leech 2004b: 66ff.) Written British English LOB F-LOB (Diff %) Spoken British English mini corpora SEU ICE-GB 16 (Diff %) Modals: must - 29.0* - 60.7* shall - 43.7* - 34.6 need 17 - 43.6* (non-existent) Semi-modals: (HAVE) got to, gotta - 34.1 - 25.7 HAVE to + 9.0 + 31.6 NEED to + 249.1* + 650.0* * significant at p > 0.01. Note: CAPITALIZED forms represent all morphological variants. Ongoing grammatical change in spoken British English 39 Leech’s comparison of LOB and F-LOB reveals that all the central modals relevant to this study appear to be decreasing in their frequency in written British English. Among the group of the semi-modals considered, there is no clear overall picture. Whereas need to is increasing drastically, the increase in have to is less pronounced. (Have) got to, on the other hand, seems to be declining. Mair (2006b: 105) synchronically analyzed the Longman Spoken American Corpus and the conversation component of ICE-GB. He found out that have (got) to is much more common than must in the present-day speech of both varieties. He however attributes this change only to some extent to the fact that have to has additional syntactically-motivated uses. Like Smith (2003: 255), Mair claims that it is in particular the non-syntactically motivated forms that are increasing. Moreover, he concludes that have to is replacing must in its deontic use, but not in epistemic contexts (Mair 2006b: 105). According to the previous studies it seems as if the two registers, though at a different pace, are heading in the same direction. Even though being closely aligned in terms of the number of words, the provisional two mini corpora used are not as closely matched as the DCPSE regarding timeframe and transcription files. The DCPSE is thus expected to allow a more in-depth examination of ongoing changes. Unlike the genitive or the subjunctive, the semi-modals are characteristic of spoken language, which leads to the assumption that innovations in their use originate in spontaneous speech, and subsequently spread to writing. My paper will be limited to the modals and semi-modals that have been reported to be most innovative. Since need to is reported to increase dramatically, the analysis of must and have (got) to is necessary, because changes in need to are very likely to affect other modals expressing the same concepts of obligation and necessity. Shall is remarkable because it has been declining for at least two hundred years, and it will be interesting to find out which uses still survive in spoken British English. Barbara Klein 40 18 In order to corroborate this assumption, an auditory analysis is needed, which is however beyond the scope of my paper. Table 5 presents the frequencies of the modals and semi-modals of obligation and necessity, including negated forms and contractions. Table 5: Frequencies of the modals and semi-modals of obligation and necessity in two sub-samples of the DCPSE DCPSE 1958-1960 1990-1992 Log likhd Diff (%) total n/ 10,000 total n/ 10,000 must 38 10.21 195 4.63 16.61** - 54.67 (HAVE) got to 24 6.45 185 4.39 2.84 - 31.90 HAVE to 34 9.31 555 13.17 4.97* + 44.21 need 0 0.00 1 0.02 0.17 (+ 100.00) NEED to 1 0.27 116 2.75 13.15** + 924.80 Total 97 26.06 1052 24.97 0.16 - 4.19 *HAVE to 1958-60 vs.1990-92: significant at p>0.05; **must, need to 1958-60 vs.1990-92: significant at p>0.01. CAPITALIZED forms represent all morphological variants. The DCPSE data clearly indicate that, in spite of the overall stability, individual frequencies of modals and semi-modals of obligation and necessity have changed considerably within the thirty years examined. In the scope of this reallocation, core modal must has suffered most, dropping by 54.67% from 10.21 to 4.63 occurrences per 10.000 words. This decrease is not only statistically significant, but also much more pronounced than the decline in must in written British English (- 29.0%). Interestingly, (have) got to, though being considered as informal and characteristically British (cf. Huddleston and Pullum 2002: 113), is declining in the DCPSE almost to the same extent as in writing from LOB to F-LOB (- 34.1%). The proportion of examples in which have is omitted is negligibly small in both corpora. In the late 1950s corpus, have is dropped in none of the examples, and even in the early 1990s, have-omission amounts only to 3.78% of all instances of got to. It can however be suspected that even though have-omission is still rare, the phonetic reduction of got to gotta is already taking place in British speech. 18 The DCPSE furthermore shows that, unlike must and (have) got to, have to is increasingly used in British English, rising by 44.21%. While the written corpora only list an increase by 9.0%, Leech’s (2004b: 69) analysis of the Ongoing grammatical change in spoken British English 41 19 Two instances of shan’t are included in the 1990s. spoken mini corpora (increase by 31.6%) has already foreshadowed the growing importance of have to. In the case of need, the trends for auxiliary and main verb uses in spoken English are highly divergent. Auxiliary need followed by a bare verb form does not occur in the DCPSE. The only instance of modal usage is the negative contraction needn’t, which however only occurs once in the year 1992. On the basis of the DCPSE, it is thus possible to argue that need as an auxiliary verb remains virtually unused in present day spoken British English. Main verb need to, however, is drastically increasing in frequency, mounting from 0.27 to 2.75 occurrences per 10.000 words, which represents an extraordinary increase by 924.80%. The rise of need to, highly significant at the 1% level, is distinctly the most pronounced development among the modals and semi-modals. It is remarkable that this form was only very rarely used in the late 1950s, and is now ranking similar to (have) got to or must. Example (6) represents a typical use of need to: (6) I just need to check your blood pressure <DI-A10_197>. Similar to core modal must, shall has declined considerably in its frequency in spoken British English. The drop by 59.80% from the late 1950s to the early 1990s is statistically significant at p > 0.01. Table 6: Frequency of modal auxiliary shall in two sub-samples of the DCPSE DCPSE 1958-1960 1990-1992 Log likhd Diff (%) total n/ 10,000 total n/ 10,000 shall 20 5.37 91 19 2.16 11.14* - 59.80 * significant at p>0.01 In the field of the modals and semi-modals of obligation and necessity, a close look at the synchronic distribution is necessary for the sake of comparing the frequencies of the individual forms. Table 7 lists the verbs must, have to, have got to, and need to in their order of precedence in written (LOB/ F-LOB) and spoken British English (DCPSE). The numbers represent the normalized frequencies per 10,000 words. Barbara Klein 42 20 The numbers for the written corpora are adapted from the results by Smith (2003: 248), who normalized the frequencies per 100,000 words. Table 7: Modals and semi-modals of obligation and necessity in their order of precedence in speech and writing LOB 20 F-LOB DCPSE 1958-60 DCPSE 1990-92 rank n/ 10,000 n/ 10,000 n/ 10,000 n/ 10,000 1 must 11.41 HAVE to 8.17 must 10.21 HAVE to 13.17 2 HAVE to 7.53 must 8.07 HAVE to 9.13 must 4.63 3 (HAVE) 4.11 NEED to 1.96 (HAVE) 6.45 (HAVE) 4.39 got to got to got to 4 NEED to 0.54 (HAVE) 0.27 NEED to 0.27 NEED to 2.75 got to Total 23.59 18.50 26.06 24.94 Must and have to have changed places in both written and spoken British English within a thirty-year window. Whether this reversal is due to a replacement of must by have to is a question which cannot be answered straightforwardly, as there is a complex interaction of semantic and syntactic factors. The synchronic comparison of F-LOB and the 1990s DCPSE part furthermore shows that have to, as was suspected by Leech (2004b: 69), is used more frequently in speech than it is in writing. In addition, (have) got to, even though declining, is still more common in the spoken language of the early 1990s than in writing. Another interesting result of this ranking is the fact that need to, even though increasing dramatically in speech, is still the least frequently used option among the four verbs. Unlike the two preceding case studies, developments in speech and writing are broadly comparable with this variable. What divides writing and speech is not the presence or absence of a development but merely the speed at which these reorganizations in the modal system are happening: spontaneous speech clearly is more innovative here, and spoken corpora should therefore be the first choice for studying changes in the modals. Of course, the statistical survey needs to be continued by qualitative follow-up studies. The first step in the semantic analysis is the classification of the examples found according to various meanings and functions of the modals. Table 8 presents such a differentiated analysis for must in the DCPSE. Ongoing grammatical change in spoken British English 43 21 <DI-B44_161> Loose shirts over jeans has got to be a sort of temporary prejudice, hasn’t it 22 From LOB to F-LOB, Smith (2003: 257) documented a significant increase of epistemic have to by 225.0%. Log likelihood 10.0; significant at p < 0.01. Table 8: Meanings of must DCPSE 1958-1960 1990-1992 Log likhd Diff (%) frequency proportion frequency proportion must epistemic 15 39.47% 93 47.69% 4.05* - 45.23 must deontic 23 60.53% 102 52.31% 13.43** - 60.82 total 38 100.00% 195 100.00% - 54.67 *must epistemic 1958-60 vs. 1990-92: significant at p>0.05; **must deontic 1958-60 vs. 1990-92: significant at p>0.01 As these data reveal, must is losing ground both in its epistemic and in its deontic function. The decline in deontic uses is slightly more pronounced. Again, the situation in written British English is similar: Smith (2003: 257) compares LOB to F-LOB, and reports a decline in both meanings of must. As in speech, the drop of epistemic must in writing is less severe (epistemic: 13.2%; deontic: 33.5%). In the case of (have) got to, the overall usage in speech is different. Apart from one single exception 21 all examples of (have) got to in the two parts of the DCPSE considered are deontic, which means that the reported decline can only be affecting its deontic uses. Since the meaning of deontic (have) got to is closer to must than to have to, the decline makes sense. Surprisingly, all DCPSE examples of have to are deontic as well. Consequently, Leech’s (2004a: 81) assumption that epistemic have to is “growing more common in British English” appears to apply to written English exclusively. 22 Since it is generally agreed that need to is rarely able to express epistemic modality, it is not surprising that it is likewise exclusively used in its deontic sense in the DCPSE. The semantic decoding of the modals and semi-modals strongly suggests that the decrease in (have) got to and in the deontic uses of must is compensated by the respective increase in have to and need to. Since must and need to cannot be used interchangeably in all contexts, and since (have) got to is not the regular alternative in past contexts, I will limit my comparison to deontic must and deontic, non-syntactically motivated have to. The major aim is to discover whether spoken British English is moving away from expressions of authority toward more polite and less pushy modes of expression. In order to retrieve only relevant and comparable Barbara Klein 44 23 According to Smith (2003: 254), these are non-finite forms, negated forms expressing absent requirement as in you don’t have to and all past forms. forms, first all epistemic uses were eliminated from the overall number. In a second step, all syntactically motivated uses 23 of have to were extracted, in order to guarantee comparability to bare verb form must. Table 9 shows the frequencies of all interchangeable forms. Table 9: Deontic must and deontic, non-syntactically motivated have to DCPSE 1958-1960 1990-1992 Log likhd Diff (n/ 10,000) Diff (%) total n/ 10,000 total n/ 10,000 deontic must 23 6.18 102 2.42 13.43** - 3.76 - 60,82 non-syntactically motivated HAVE to 22 5.91 392 9.30 4.92* + 3.39 + 57.41 *HAVE to 1958-60 vs.1990-92: significant at p>0.05; **must 1958-60 vs.1990-92: significant at p>0.01 Capitalized HAVE here represents the present tense forms have and has. According to these data, it looks as if the increase in non-syntactically motivated have to, which is by definition in the majority of cases interchangeable with must, compensates for the decrease in deontic must. As we have seen before, must is not threatened by have to in epistemic function in spoken English. In regular finite, non-negated present tense, it appears as if have to was becoming more and more popular in present-day British English speech. Statements expressing obligation by the use of have to such as (7) are definitely widely spread today: (7) Oh you have to pay for these <DI-B25_50> However, it is absolutely essential to point out that it would be premature to deduce that there is a one-to-one replacement of must by have to. It is obvious that need to is also replacing must in some deontic contexts in the same way in which (have) got to can also be used instead of have to. The case of declining shall is different from the changes in the modals and semi-modals discussed so far, because over the centuries, some of its uses have gradually disappeared (cf. Gotti 2003: 278f.). Table 10 shows the development of the two main remaining usages in the second part of the twentieth century: Ongoing grammatical change in spoken British English 45 Table 10: Meanings of shall DCPSE 1958-1960 1990-1992 Log likhd Diff (%) frequency proportion frequency proportion shall volition 4 20.00% 43 47.25% 0.01 - 5.03 shall prediction 16 80.00% 48 52.75% 16.51* - 73.50 total 20 100.00% 91 100.00% 11.14* - 59.80 *shall prediction and total 1958-60 vs.1990-92: significant at p>0.01 According to the DCPSE data, shall has been losing ground both in expressing volition and prediction. The decline in its futurity meaning is clearly more distinct, which is probably due to modal will becoming more popular and replacing shall in future contexts. What is striking is the high number of “formulaic discourse structuring devices” (Mair 2006b: 102) of the type as we shall see, which are not productive, and thus hint at the fact that shall might become obsolete in the course of the following centuries. Since they encompass both the spoken and the written language, the developments in the field of modality probably reflect more genuinely grammatical changes than the shifts in the use of the genitive and the subjunctive. With the “emerging semi-modals” (cf. the title of Krug 2000) have to and need to, we are seeing grammaticalization at work. Colloquialization is “a tendency for features of conversational spoken language to infiltrate, and spread in the written language” (Leech 2004b: 75). Thereby, have to and need to provide prime examples. Their rate of increase in speech is three times that in writing, which suggests that the trend originated in spontaneous speech and subsequently spread to writing, where it is thus still lagging behind somewhat. Democratization, a “tendency to avoid unequal and face-threatening modes of interaction” (Leech 2004b: 75) is a socio-psychological model of language change, which can be perfectly applied to the decline in must and the respective rise in have to and need to. Myhill (1995), investigating the functions of a selection of modals before and after the American Civil War, argues that the “old” modals must, should, may, and shall were associated with hierarchical social relationships, with people controlling the actions of other people, and with absolute judgements based upon social decorum, principle and rules about societal expectations of certain types of people. The ‘new’ modals [i.e. have to, got to] are more personal, being used to give advice to an equal, make an emotional request, offer help or criticize one’s interlocutor (Myhill 1995: 157). Barbara Klein 46 Even though my paper is neither focused on American English nor on the time around the Civil War, Myhill’s description of the modals perfectly fits the trends observable in the DCPSE in the second half of the twentieth century. It is highly likely that the decline in must is the reflection of a societal tendency arising after World War II (cf. Mair and Hundt 1997: 78), which is characterized by a movement away from exercising authority openly and encourages more polite and egalitarian ways of formulating commands, orders or requests (see example 6). I conclude that in this area, changes in speech can be definitely considered prior to writing. Unlike the genitive and the subjunctive, the modals and semi-modals turned out to be not a mere stylistic option in writing, but a category of verbs whose origin lies in what the DCPSE represents: spontaneous spoken language. 5. Conclusion On the basis of the DCPSE corpus, this paper has identified and analyzed recent developments in the use of three grammatical variables within the span of one generation. They have turned out to be fairly heterogeneous both in the scope and in the potential causes of the observed changes. The genitive represents an area in which there is no significant ongoing change in present-day spoken English. Contrary to expectations raised by studies of written English, the frequencies of both s-genitives and equivalent of-constructions have remained stable within the thirty years documented in the DCPSE. Similarly, the use of the subjunctive has turned out to be more stable in speech than in writing. In speech, the present subjunctive is close to non-existent in British English. In mandative statements, the indicative is gaining ground, with even the should option declining. With modals and semi-modals, on the other hand, the diachronic dynamic seems to be greater in speech, with the changes originating there eventually being taken up in writing. This paper therefore provides evidence that one should not act on the general assumption that changes in speech always precede changes in writing. It appears as though changes in formal or stylistically optional grammatical elements in writing may progress independently of the spoken register. Beyond the immediate focus on three grammatical variables, the present paper can also be seen as a test for a new corpus resource, the DCPSE. In the compilation of this corpus, considerable efforts have been made by the compilers to control all variables in aligning the “older” London-Lund components with the “newer” ICE-GB ones, which has generally been successful. But some shortcomings remain. One is the number of years covered in the two parts of the corpus: while the 400,000 “new” words from ICE-GB Ongoing grammatical change in spoken British English 47 cover three years, the “old” LUND part is spread over nineteen years. Since the first priority of my study was an exact match of years in order to guarantee comparability to the written corpora, I had to work with corpus parts of different sizes. As a consequence, text categories could not be completely matched. This study is one of the first to be carried out on the basis of the DCPSE, and there will be more to follow. The field of phonetics will gain from this new corpus as well, since auditory files of the original recordings are included. In the vast field of morphosyntax, this paper can be considered as a mere starting point. As we have seen, the nature of the changes in speech and writing can vary considerably, which calls for further comparisons of the two registers using the DCPSE. In conclusion, it is imperative to emphasize the need for comparable corpora of spoken American English to create a set of spoken corpora similar to LOB and Brown, and their respective updates. In many ways, an American counterpart to the DCPSE would be the last piece of the puzzle, completing and uniting the existing corpora, and serving as a unique basis for an integrated study of regional variability and diachronic change in the recent history of spoken English. References Algeo, John (1988). “British and American grammatical differences.” International Journal of Lexicography 1. 1-31. Barber, Charles (1964). Linguistic Change in Present-Day English. London & Edinburgh: Oliver and Boyd. Dahl, Liisa (1971). “The s-genitive with non-personal nouns in modern English journalistic style.” Neuphilologische Mitteilungen 72. 140-172. Fowler, Henry W (1926). A Dictionary of Modern English Usage. London: Milford. Gotti, Maurizio (2003). “Shall and will in contemporary English: a comparison with past uses.” In: Roberta Facchinetti, Manfred Krug, and Frank Palmer (eds.).Modality in Contemporary English. Berlin & New York: Mouton de Gruyter. 267-300. Hinrichs, Lars and Benedikt Szmrecsanyi (2007). “Recent changes in the function and frequency of Standard English genitive constructions: a multivariate analysis of tagged corpora.” English Language and Linguistics 11: 3. 437-474. Hirtle, Walter H (1964). “The English present subjunctive.” Canadian Journal of Linguistics 9. 75-82. Huddleston, Rodney and Geoffrey K. Pullum (2002). The Cambridge Grammar of the English Language. Cambridge: Cambridge UP. Hundt, Marianne (1998). “It is important that this study (should) be based on the analysis of parallel corpora: On the use of the mandative subjunctive in four major varieties of English.” In: Hans Lindquist et al. (eds.).The Major Varieties of English. Papers from MAVEN 97. Växjö: Acta Wexionensia. 159-175. Jacobsson, Bengt (1975). “How dead is the English subjunctive.” Moderna Språk 69. 218-231. Barbara Klein 48 Jespersen, Otto (1905). Growth and Structure of the English Language. Chicago: Chicago UP. Johansson, Stig and Else Helene Norheim (1988). “The subjunctive in British and American English.” ICAME Journal 12. 27-36. Kreyer, Rolf (2003). “Genitive and of-construction in modern written English. Processability and human involvement.” International Journal of Corpus Linguistics 8. 169-207. Krug, Manfred (2000). Emerging English Modals: A Corpus-Based Study of Grammaticalization. Berlin/ New York: Mouton de Gruyter. Leech, Geoffrey (2003). “Modality on the move: The English modal auxiliaries 1961-1992.” In: Roberta Facchinetti, Manfred Krug, and Frank Palmer (eds.).Modality in Contemporary English. Berlin & New York: Mouton de Gruyter. 223-240. Leech, Geoffrey (2004a). Meaning and the English Verb. Harlow: Longman/ Pearson. Leech, Geoffrey (2004b). “Recent grammatical change in English: data, description, theory.” In: Bengt Altenberg and Karin Ajimer (eds.).Advances in corpus linguistics. Proceedings of the 23 rd ICAME Conference, Gothenburg, 2002. Amsterdam & New York: Rodopi. 61-81. Leech, Geoffrey and Nicolas Smith (2006). “Recent grammatical change in written English 1961-1992: some preliminary findings of a comparison of American with British English.” In: Antoinette Renouf and Andrew Kehoe (eds.).The Changing Face of Corpus Linguistics. Amsterdam & New York: Rodopi. 185-204. Mair, Christian und Marianne Hundt (1997). “The Corpus-Based Approach to Language Change in Progress.” In: Uwe Böker and Hans Sauer (eds.).Anglistentag 1996 Dresden. Trier: WVT. 71-82. Mair, Christian, Marianne Hundt, Geoffrey Leech and Nicholas Smith (2002). “Short-Term Diachronic Shifts in Part-of-Speech Frequencies: A Comparison of the Tagged LOB and F-LOB Corpora.” International Journal of Corpus Linguistics 7. 245-264. Mair, Christian (2006a). “Inflected genitives are spreading in present-day English, but not necessarily to inanimate nouns.” In: Christian Mair and Reinhard Heuberger (eds.). Corpora and the History of English: Festschrift für Manfred Markus. Heidelberg: Winter. 235-248. Mair, Christian (2006b). Twentieth Century English. History, Variation and Standardization. Cambridge: Cambridge UP. Mair, Christian and Geoffrey Leech (2006). “Current Changes in English Syntax.” In: Bas Aarts and April McMahon (eds.).The Handbook of English Linguistics. Oxford: Blackwell. 318-342. Myhill, John (1995). “Change and continuity in the functions of the American English modals.” Linguistics 33. 157-211. Nichols, Ann Eljenholm (1987). “The suasive subjunctive: Alive and well in the upper Midwest.” American Speech 62. 140-153. Övergaard, Gerd (1995). The mandative subjunctive in American and British English. Stockholm: Almquist & Wiksell. Potter, Simeon (1969, 2 nd ed. 1975). Changing English. London: Deutsch. Raab-Fischer, Roswitha (1995). “Löst der Genitiv die of-Phrase ab? Eine korpusgestützte Studie zum Sprachwandel im heutigen Englisch.” Zeitschrift für Anglistik und Amerikanistik 43. 123-132. Rosenbach, Anette (2002). Genitive Variation in English: Conceptual Factors in Synchronic and Diachronic Studies. Berlin: Mouton de Gruyter. Ongoing grammatical change in spoken British English 49 Serpollet, Noëlle (2001). “The mandative subjunctive in British English seems to be alive and kicking…Is this due to the influence of American English? ” In: Paul Rayson, Andrew Wilson, Tony McEnery Andrew Hardie, and Shereen Khoja (eds.).Proceedings of the Corpus Linguistics 2001 conference. Lancaster University: UCREL Technical Papers, vol. 13. 531-542. Smith, Nicholas (2003). “Changes in the modals and semi-modals of strong obligation and epistemic necessity in recent British English.” In: Roberta Facchinetti, Manfred Krug, and Frank Palmer (eds.).Modality in Contemporary English. Berlin & New York: Mouton de Gruyter. 241-266. Szmrecsanyi, Benedikt and Lars Hinrichs (2008). “Probabilistic determinants of genitive variation in spoken and written English: a multivariate comparison across time, space, and genres.” In: Terttu Nevalainen, Irma Taavitsainen, Päivi Pahta, and Minna Korhonen (eds.).The Dynamics of Linguistic Variation: Corpus Evidence on English Past and Present. Amsterdam: Benjamins. Wallis, Sean and Bas Aarts. The Diachronic Corpus of Present-Day Spoken English (DCPSE). CD-ROM. London: Survey of English Usage, UCL. Barbara Klein Englisches Seminar Universität Freiburg